You are on page 1of 13

Designation: D 2915 03

Standard Practice for


Evaluating Allowable Properties for Grades of Structural
Lumber1
This standard is issued under the fixed designation D 2915; the number immediately following the designation indicates the year of
original adoption or, in the case of revision, the year of last revision. A number in parentheses indicates the year of last reapproval. A
superscript epsilon (e) indicates an editorial change since the last revision or reapproval.

INTRODUCTION

The mechanical properties of structural lumber depend upon natural growth characteristics and
manufacturing practices. Several procedures can be used to sort lumber into property classes or stress
grades, the most widely used being the visual methods outlined in Practice D 245. With each, a
modulus of elasticity and a set of from one to five allowable stresses may be associated with each
stress grade. The allowable stresses are extreme fiber stress in bending, tension parallel to the grain,
compression parallel to the grain, shear, and compression perpendicular to the grain. This test method
for evaluation of the properties of structural lumber defines an allowable property as the value of the
property that would normally be published with the grade description.
This practice is useful in assessing the appropriateness of the assigned properties and for checking
the effectiveness of grading procedures.
For situations where a manufactured product is sampled repeatedly or lot sizes are small, alternative
test methods as described in Ref (1)2 may be more applicable.

1. Scope 1.2 The values stated in inch-pound units are to be regarded


1.1 This practice covers sampling and analysis procedures as the standard.
for the investigation of specified populations of stress-graded 1.3 This standard does not purport to address all of the
structural lumber. Depending on the interest of the user, the safety concerns, if any, associated with its use. It is the
population from which samples are taken may range from the responsibility of the user of this standard to establish appro-
lumber from a specific mill to all the lumber produced in a priate safety and health practices and determine the applica-
particular grade from a particular geographic area, during some bility of regulatory limitations prior to use.
specified interval of time. This practice generally assumes that 2. Referenced Documents
the population is sufficiently large so that, for sampling
purposes, it may be considered infinite. Where this assumption 2.1 ASTM Standards:
is inadequate, that is, the population is assumed finite, many of D 198 Test Methods of Static Tests of Timber in Structural
the provisions of this practice may be employed but the Sizes3
sampling and analysis procedure must be designed to reflect a D 245 Practice for Establishing Structural Grades and Re-
finite population. The statistical techniques embodied in this lated Allowable Properties for Visually Graded Lumber3
practice provide procedures to summarize data so that logical D 1990 Practice for Establishing Allowable Properties for
judgments can be made. This practice does not specify the Visually-Graded Dimension Lumber from In-Grade Tests
action to be taken after the results have been analyzed. The of Full-Size Specimens3
action to be taken depends on the particular requirements of the E 105 Practice for Probability Sampling of Materials4
user of the product. 3. Statistical Methodology
3.1 Two general analysis procedures are described under
1
This practice is under the jurisdiction of ASTM Committee D07 on Wood and this practice, parametric and nonparametric. The parametric
is the direct responsibility of Subcommittee D07.02 on Lumber and Engineered approach assumes a known distribution of the underlying
Wood Products.
Current edition approved April 10, 2003. Published June 2003. Originally
population, an assumption which, if incorrect, may lead to
approved in 1970 as D 2915 70 T. Last previous edition approved in 2002 as
D 2915 02.
2 3
The boldface numbers in parentheses refer to the list of references at the end of Annual Book of ASTM Standards, Vol 04.10.
4
this practice. Annual Book of ASTM Standards, Vol 14.02.

Copyright ASTM International, 100 Barr Harbor Drive, PO Box C700, West Conshohocken, PA 19428-2959, United States.

1
D 2915 03
inaccurate results. Therefore, if a parametric approach is used, may be estimated. For all properties, nonparametric or para-
appropriate statistical tests shall be employed to substantiate metric techniques are applicable. Commonly the mean modu-
this choice along with measures of test adequacy (2, 3, 4, 5, 6, lus of elasticity and the mean compression perpendicular to the
7). Alternatively a nonparametric approach requires fewer grain stress for the grade are estimated. For the four other
assumptions, and is generally more conservative than a para- allowable stresses, a near-minimum property is generally the
metric procedure. objective.
3.2 Population: 3.4.2 Determine sample size sufficient for estimating the
3.2.1 It is imperative that the population to be evaluated be mean by a two-stage method, with the use of the following
clearly defined, as inferences made pertain only to that popu- equation. This equation assumes the data is normally distrib-
lation. In order to define the population, it may be necessary to uted and the mean is to be estimated to within 5 % with
specify ( 1) grade name and description, (2) geographical area specified confidence:
over which sampling will take place (nation, state, mill, etc.),
(3) species or species group, (4) time span for sampling (a S t
n 5 ~ts/0.05 X! 2 5 0.05 CV D 2
(1)
days production, a month, a year, etc.), (5) lumber size, and
(6) moisture content. where:
n = sample size,
3.2.2 Where possible, the sampling program should con- s = standard deviation of specimen values,
sider the location and type of log source from which the pieces X = specimen mean value,
originated, including types of processing methods or marketing CV = coefficient of variation, s/ X,
practices with respect to any influence they may have on the 0.05 = precision of estimate, and
representative nature of the sample. Samples may be collected t = value of the t statistic from Table 1.
from stock at mills, centers of distribution, at points of end use Often the values of s, X, and t or CV and t are not known
or directly from current production at the grading chains of before the testing program begins. However, s and X, or CV,
manufacturing facilities. may be approximated by using the results of some other test
3.3 Sampling Procedure: program, or they may simply be guessed (see example, Note
3.3.1 Random Sampling The sampling unit is commonly 1).
the individual piece of lumber. When this is not the case, see NOTE 1An example of initial sample size calculation is:
3.3.3. The sampling shall assure random selection of sampling Sampling a grade of lumber for modulus of elasticity (E). Assuming a
units from the population described in 3.2 with all members of 95 % confidence level, the t statistic can be approximated by 2.
the population sharing equal probability of selection. The s = 300 000 psi (2067 MPa)
principles of Practice E 105 shall be maintained. When sam- X = assigned E of the grade = 1 800 000 psi (12 402 MPa)
pling current production, refer to Practice E 105 for a recom- CV = (300 000/1 800 000) = 0.167
t = 2
mended sampling procedure (see Appendix X3 of this practice
for an example of this procedure). If samples are selected from n = S 2
0.05 3 0.167 D 2
5 44.622 ~45 pieces!
inventory, random number tables may be used to determine Calculate the sample mean and standard deviation and use them to
which pieces will be taken for the sample. estimate a new sample size from Eq 1, where the value of t is taken from
3.3.2 Sampling with Unequal ProbabilitiesUnder some Table 1. If the second sample size exceeds the first, the first sample was
insufficient; obtain and test the additional specimens.
circumstances, it may be advisable to sample with unequal but
NOTE 2More details of this two-stage method are given in Ref (8).
known probabilities. Where this is done, the general principles
of Practice E 105 shall be maintained, and the sampling 3.4.3 To determine sample size based on a tolerance limit
method shall be completely reported. (TL), the desired content (C) (Note 3) and associated confi-
3.3.3 Sequential SamplingWhen trying to characterize dence level must be selected. The choice of a specified content
how a certain population of lumber may perform in a structure, and confidence is dependent upon the end-use of the material,
it may be deemed more appropriate to choose a sampling unit, economic considerations, current design practices, code re-
such as a package, that is more representative of how the quirements, etc. For example, a content of 95 % and a
lumber will be selected for use. Such a composite sampling confidence level of 75 % may be appropriate for a specific
unit might consist of a sequential series of pieces chosen to property of structural lumber. Different confidence levels may
permit estimation of the properties of the unit as well as the be suitable for different products or specific end uses. Appro-
pieces. Where this is done, the principles in 3.3.1 and 3.3.2 priate content and confidence levels shall be selected before the
apply to these composite sampling units and the sampling sampling plan is designed.
method shall be completely reported. NOTE 3The content, C, is an estimate of the proportion of the
3.4 Sample Size: population that lies above the tolerance limit. For example, a tolerance
3.4.1 Selection of a sample size depends upon the property limit with a content of 95 % describes a level at which 95 % of the
or properties to be estimated, the actual variation in properties population lies above the tolerance limit. The confidence with which this
inference is to be made is a separate statement.
occurring in the population, and the precision with which the
property is to be estimated. For the five allowable stresses and 3.4.3.1 To determine the sample size for near-minimum
the modulus of elasticity various percentiles of the population properties, the nonparametric tolerance limit concept of Ref (8)

2
D 2915 03
TABLE 1 Values of the t Statistics Used in Calculating TABLE 2 Sample Size and Order Statistic for Estimating the 5 %
Confidence IntervalsA Nonparametric Tolerance Limit, NTLA
df 75 % Confidence 95 % Confidence 99 % confidence
CI = 75 % CI = 95 % CI = 99 %
n1
Sample Order Sample Order Sample Order
1 2.414 12.706 63.657 SizeB StatisticC Size Statistic Size Statistic
2 1.604 4.303 9.925
28 1 59 1 90 1
3 1.423 3.182 5.841
53 2 93 2 130 2
4 1.344 2.776 4.604
78 3 124 3 165 3
5 1.301 2.571 4.032
102 4 153 4 198 4
125 5 181 5 229 5
6 1.273 2.447 3.707
148 6 208 6 259 6
7 1.254 2.365 3.499
170 7 234 7 288 7
8 1.240 2.306 3.355
193 8 260 8 316 8
9 1.230 2.262 3.250
215 9 286 9 344 9
10 1.221 2.228 3.169
237 10 311 10 371 10
259 11 336 11 398 11
11 1.214 2.201 3.106
281 12 361 12 425 12
12 1.209 2.179 3.055
303 13 386 13 451 13
13 1.204 2.160 3.012
325 14 410 14 478 14
14 1.200 2.145 2.977
347 15 434 15 504 15
15 1.197 2.131 2.947
455 20 554 20 631 20
562 25 671 25 755 25
16 1.194 2.120 2.921
668 30 786 30 877 30
17 1.191 2.110 2.898
879 40 1013 40 1115 40
18 1.189 2.101 2.878
1089 50 1237 50 1349 50
19 1.187 2.093 2.861
A
20 1.185 2.086 2.845 Adapted from Ref (12). For other tolerance limits or confidence levels, see Ref
(12) or (8).
B
21 1.183 2.080 2.831 Where the sample size falls between two order statistics (for example, 27 and
22 1.182 2.074 2.891 28 for the first order statistic at 75 confidence), the larger of the two is shown in the
23 1.180 2.069 2.807 table, and the confidence is greater than the nominal value.
C
24 1.179 2.064 2.797 The rank of the ordered observations, beginning with the smallest.
25 1.178 2.060 2.787

26 1.177 2.056 2.779


27
28
29
1.176
1.175
1.174
2.052
2.048
2.045
2.771
2.763
2.756
SE 5 s 1 K2
n 1 2~n 2 1! (2)
30 1.173 2.042 2.750
where:
40 1.167 2.021 2.704 s = standard deviation of specimen values,
60 1.162 2.000 2.660
120 1.156 1.980 2.617
n = sample size, and
` 1.150 1.960 2.576 K = confidence level factor.
A
Adapted from Ref (8). For calculating other confidence levels, see Ref (8). The sample size, n, may be chosen to make this quantity
sufficiently small for the intended end use of the material (Note
4).
may be used (Table 2). This will provide the sample size NOTE 4An example of sample size calculation where the purpose is
suitable for several options in subsequent near-minimum to estimate a near minimum property is shown in the following calcula-
analyses. Although the frequency with which the tolerance tion:
Estimate the sample size, n, for a compression parallel strength test in
limit will fall above (or below) the population value, corre-
which normality will be assumed. A CV of 22 % and a mean C11 of 4600
sponding to the required content, is controlled by the confi- psi are assumed based on other tests. The target PTL of the lumber grade
dence level selected, the larger the sample size the more likely is 2700 psi. The PTL is to be estimated with a content of 95 % (5 % PTL)
the tolerance limit will be close to the population value. It is, and a confidence of 75 %.
therefore, desirable to select a sample size as large as possible CV = 0.22
commensurate with the cost of sampling and testing (see also X = 4600 psi (31.7 MPa)
s = (0.22) (4600) = 1012 psi (7.0 MPa)
4.7). K = ( X PTL)/s = 1.877
3.4.3.2 If a parametric approach is used, then a tolerance From Table 3:
limit with stated content and confidence can be obtained for K = 1.869 for n = 30
any sample size; however, the limitation expressed in 3.4.3.1 Therefore n ' 30 specimens.
applies. That is, although the frequency that the tolerance limit
falls above (or below) the population value, corresponding to SE 5 1012 1
1
1.8772
30 2~30 2 1! (3)
the required content is controlled, the probability that the
5 310.5 psi ~2.1 MPa!
tolerance limit will be close to the population value depends on
Consequently, although 30 specimens is sufficient to estimate the 5 %
the sample size. For example, if normality is assumed, the
PTL with 75 % confidence, the standard error (approximately 12 % of the
parametric tolerance limit (PTL) will be of the form PTL = X PTL) illustrates that, with this size sample, the PTL estimated by test may
Ks, (see Ref (8)), and the standard error (SE) of this statistic not be as close to the true population fifth percentile as desired. A larger
may be approximated by the following equation: n may be desirable.

3
D 2915 03
TABLE 3 K Factors for One-Sided Tolerance Limits for Normal DistributionsA
75 % Confidence (g = 0.25) 95 % Confidence (g = 0.05) 99 % Confidence (g = 0.01)
1p 0.75 0.90 0.95 0.99 0.75 0.90 0.95 0.99 0.75 0.90 0.95 0.99
n
3 1.464 2.501 3.152 4.397 3.805 6.156 7.657 10.555 8.726 13.997 17.374 23.900
4 1.255 2.134 2.681 3.726 2.617 4.162 5.145 7.044 4.714 7.381 9.085 12.389
5 1.151 1.962 2.464 3.422 2.149 3.407 4.203 5.742 3.453 5.362 6.580 8.941

6 1.087 1.859 2.336 3.244 1.895 3.007 3.708 5.063 2.847 4.412 5.407 7.336
7 1.043 1.790 2.251 3.127 1.732 2.756 3.400 4.643 2.490 3.860 4.729 6.413
8 1.010 1.740 2.189 3.042 1.617 2.582 3.188 4.355 2.253 3.498 4.286 5.813
9 0.984 1.702 2.142 2.978 1.532 2.454 3.032 4.144 2.083 3.241 3.973 5.390
10 0.964 1.671 2.104 2.927 1.465 2.355 2.912 3.982 1.954 3.048 3.739 5.075

11 0.946 1.646 2.074 2.886 1.411 2.276 2.816 3.853 1.852 2.898 3.557 4.830
12 0.932 1.625 2.048 2.852 1.366 2.210 2.737 3.748 1.770 2.777 3.411 4.634
13 0.919 1.607 2.026 2.823 1.328 2.156 2.671 3.660 1.702 2.677 3.290 4.473
14 0.908 1.591 2.008 2.797 1.296 2.109 2.615 3.585 1.644 2.593 3.189 4.338
15 0.899 1.577 1.991 2.776 1.267 2.069 2.566 3.521 1.595 2.522 3.103 4.223

16 0.890 1.565 1.977 2.756 1.242 2.033 2.524 3.465 1.552 2.460 3.028 4.124
17 0.883 1.555 1.964 2.739 1.220 2.002 2.487 3.415 1.514 2.405 2.963 4.037
18 0.876 1.545 1.952 2.724 1.200 1.974 2.453 3.371 1.480 2.357 2.906 3.961
19 0.869 1.536 1.942 2.710 1.182 1.949 2.424 3.331 1.450 2.314 2.854 3.893
20 0.864 1.528 1.932 2.697 1.166 1.926 2.396 3.296 1.423 2.276 2.808 3.832

21 0.858 1.521 1.924 2.686 1.151 1.906 2.372 3.263 1.398 2.241 2.767 3.777
22 0.854 1.514 1.916 2.675 1.138 1.887 2.349 3.234 1.376 2.209 2.729 3.727
23 0.849 1.508 1.908 2.666 1.125 1.869 2.329 3.207 1.355 2.180 2.695 3.682
24 0.845 1.502 1.901 2.657 1.113 1.853 2.310 3.182 1.336 2.154 2.663 3.640
25 0.841 1.497 1.895 2.648 1.103 1.838 2.292 3.159 1.319 2.129 2.634 3.602

30 0.825 1.475 1.869 2.614 1.058 1.778 2.220 3.064 1.247 2.030 2.516 3.447
35 0.812 1.458 1.849 2.588 1.025 1.732 2.167 2.995 1.194 1.958 2.430 3.335
40 0.802 1.445 1.834 2.568 0.999 1.697 2.126 2.941 1.154 1.902 2.365 3.249
45 0.794 1.434 1.822 2.552 0.978 1.669 2.093 2.898 1.121 1.857 2.312 3.181
50 0.788 1.426 1.811 2.539 0.960 1.646 2.065 2.863 1.094 1.821 2.269 3.125

60 0.777 1.412 1.795 2.518 0.932 1.609 2.023 2.808 1.051 1.764 2.203 3.039
70 0.769 1.401 1.783 2.502 0.911 1.581 1.990 2.766 1.019 1.722 2.153 2.974
80 0.762 1.393 1.773 2.489 0.894 1.560 1.965 2.733 0.994 1.689 2.114 2.924
90 0.757 1.386 1.765 2.479 0.881 1.542 1.944 2.707 0.974 1.662 2.083 2.884
100 0.753 1.380 1.758 2.470 0.869 1.527 1.927 2.684 0.957 1.639 2.057 2.850

120 0.745 1.371 1.747 2.456 0.851 1.503 1.900 2.650 0.930 1.604 2.016 2.797
140 0.740 1.364 1.739 2.446 0.837 1.485 1.879 2.623 0.909 1.577 1.985 2.758
160 0.736 1.358 1.733 2.438 0.826 1.471 1.862 2.602 0.893 1.556 1.960 2.726
180 0.732 1.353 1.727 2.431 0.817 1.460 1.849 2.585 0.879 1.539 1.940 2.700
200 0.729 1.350 1.723 2.425 0.809 1.450 1.838 2.570 0.868 1.524 1.923 2.679

250 0.723 1.342 1.714 2.414 0.794 1.431 1.816 2.542 0.846 1.496 1.891 2.638
300 0.719 1.337 1.708 2.406 0.783 1.417 1.800 2.522 0.830 1.476 1.868 2.609
350 0.715 1.332 1.703 2.400 0.775 1.407 1.788 2.507 0.818 1.461 1.850 2.586
400 0.712 1.329 1.699 2.395 0.768 1.398 1.778 2.495 0.809 1.449 1.836 2.568
450 0.710 1.326 1.696 2.391 0.763 1.391 1.770 2.484 0.801 1.438 1.824 2.553

500 0.708 1.324 1.693 2.387 0.758 1.385 1.763 2.476 0.794 1.430 1.815 2.541
600 0.705 1.320 1.689 2.382 0.750 1.376 1.753 2.462 0.783 1.416 1.799 2.521
700 0.703 1.317 1.686 2.378 0.745 1.369 1.744 2.452 0.775 1.406 1.787 2.506
800 0.701 1.315 1.683 2.374 0.740 1.363 1.738 2.443 0.768 1.398 1.777 2.493
900 0.699 1.313 1.681 2.371 0.736 1.358 1.732 2.436 0.762 1.391 1.769 2.483

1000 0.698 1.311 1.679 2.369 0.733 1.354 1.728 2.431 0.758 1.385 1.763 2.475
1500 0.694 1.306 1.672 2.361 0.722 1.340 1.712 2.411 0.742 1.365 1.741 2.447
2000 0.691 1.302 1.669 2.356B 0.715 1.332 1.703 2.400B 0.733 1.354 1.727 2.431B
2500 0.689 1.300B 1.666B 2.353B 0.711 1.326 1.697B 2.392B 0.727 1.346 1.719B 2.419B
3000 0.688 1.299B 1.664B 2.351B 0.708 1.323B 1.692B 2.386B 0.722 1.340B 1.712B 2.411B
inf 0.674 1.282 1.645 2.326 0.674 1.282 1.645 2.326 0.674 1.282 1.645 2.326
A
Obtained from a noncentral t inverse approach; see Ref (15).
B
Computed using formula X5.2.

4
D 2915 03
3.4.4 Often the objective of the evaluation program will be TABLE 4 K Factors for Adjusting Apparent Modulus of Elasticity
to estimate mean and near-minimum properties simultaneously. of Simply Supported BeamsA
When this is the case, only one sample size need be used. It Loading Deflection measured at Ki
should be the greater of the two obtained in accordance with Concentrated at midspan midspan 1.200
3.4.2 and 3.4.3. Concentrated at third points midspan 0.939
Concentrated at third points load points 1.080
3.4.5 If a sampling unit other than an individual piece of Concentrated at outer quarter-points midspan 0.873
lumber is to be used, as provided for in 3.3.3, then the required Concentrated at outer quarter-points load points 1.20
Uniformly distributed midspan 0.960
sample size must be determined by procedures that are
A
statistically appropriate for the sampling method chosen. In the See Appendix X4 for an example of use of Table 4.
case of multisource data, as in the sampling of some or all mills
in a defined region, special procedures may be required, for
example, those based on the methodology introduced in Ref of shear deflection on beam deflection. Factors to adjust Eai for
(9). In all cases, the procedures shall be fully described. span-depth ratio and load configuration may be derived from
Eq 4, (Ref (11)). To determine the apparent modulus of
4. Analysis and Presentation of Results elasticity, Eai2, based on any set of conditions of span-depth
4.1 The results of the tests performed in accordance with ratio and load configuration, when the modulus, E ai, based on
Test Methods D 198 or other standard testing procedures shall some other set of conditions is known, solve the equation:
be analyzed and presented as (1) a set of summarizing
statistics, and (2) an appendix of unadjusted individual test
Eai2 5
h1
1 1 K1 L
1
S DS D 2 E
G
Eai (4)
specimen results. If parametric procedures are to be used, a
description of the selection procedures and a tabulation of
h2
1 1 K2 L
2
S DS D 2 E
G
distribution parameters shall be provided. Any best-fit judg- where:
ment (Note 5) between competing distributions shall be docu- h = depth of the beam,
mented. L = total beam span between supports,
NOTE 5A best-fit procedure should recognize the low power of some E = shear free modulus of elasticity,
published procedures. To check the fit, the series of tests outlined in Ref G = modulus of rigidity, and
(10) represents several alternatives. Also, tests based on the Anderson- Ki = values are given in Table 4.
Darling statistic (2, 3, 4) have been shown to be among the more powerful The equations were derived using simple beam theory for a
tests (6, 7). It should be noted, however, that not all tests are valid for all simply supported beam composed of isotropic, homogeneous
distributions and that these procedures are effective for checking central material. Experimental evidence suggests that these equations
tendency. For instance, revised standard tables of the Kolmogorov-
Smirnov statistic are presently available only for the normal, logistic, and
produce reasonable results with solid wood when converting
exponential distributions (5). between load conditions at a fixed span-depth ratio. Care must
be exercised when converting between different span-depth
4.2 Properties shall be adjusted to a single moisture content ratios to assure that the adjustments are appropriate for the end
appropriate for the objective of the testing program. Although use.
test results can be adjusted for moisture content, these adjust- 4.3.1 Often, lumber is not homogeneous within a piece with
ments decrease in accuracy with increasing change in moisture respect to modulus of elasticity. The apparent modulus, there-
content. For this reason, it is suggested that the specimens be fore, may be affected by the location of growth characteristics,
conditioned as closely as possible to the target moisture content such as knots, with respect to loads and supports. It is further
prior to test, and that adjustments for more than five percentage cautioned that conversions may be less appropriate when
points of moisture content are to be avoided. When adjust- converting between edgewise and flatwise specimen orienta-
ments are required, the procedures given in Practice D 1990 tions.
shall be used for dimensions, bending strength, modulus of 4.3.2 If modulus of elasticity results are not measured at
elasticity, tensile strength, and compressive strength parallel to standardized conditions, separate justification shall be provided
the grain. When adjusting shear strength and compressive for factors used to adjust test values to standardized conditions.
strength perpendicular to the grain the procedures of Practice 4.3.3 In calculations using Eq 4 and that involve mean
D 245 shall be used. trends of large populations, a single E/G ratio is usually
4.3 Modulus of elasticity values of primary concern are assumed.7 If this assumption is critical for the intended
apparent values, Eai, used in deflection equations that attribute
all deflection to moment. These apparent moduli may be
standardized for a specific span-depth ratio and load configu-
ration. Standardization should reflect, as far as possible, 6
A uniform load distribution is commonly encountered in use. This load
conditions of anticipated end use.5,6 When tests at standardized configuration is difficult to apply in testing, but may be closely approximated by
applying the load at the one-third points of the span, if the span-to-depth ratio is the
conditions of load and span are not possible, to adjust Eai to same.
standardized conditions, it is necessary to account for the effect 7
When using these conversion equations with solid wood, historically it has been
assumed that the modulus of rigidity (G) is one sixteenth the shear free modulus of
elasticity ( E). Limited data indicate the ratio of E/G for individual pieces of lumber
can vary significantly from this value depending upon the number, size, and location
5
Spans, which customarily serve as a basis for design range, go from 17 to 21 of knots present, the slope of grain in the piece, and the spans over which deflections
times the depth of the specimen. are measured (13).

5
D 2915 03
application, it is recommended that the moduli of elasticity and
rigidity of the individual pieces be measured (see Methods
k
F
NPE 5 100 ~n 1 1! 2 ~j 2 1! G [x j 2 x ~j 2 1!# 1 x ~j 2 1! (8)

D 198). where k is the desired percentile point estimate sought.


4.4 The adjustment factors used to reduce the test statistics
NOTE 6Order statistics are ranked test values from the lowest to the
to the level of allowable properties depend on the property and highest. For example, the first order statistic is the lowest test value or the
are shown in Table 5. They are taken from Practice D 245, weakest piece in the sample, the second order statistic is the second
which includes a safety factor and a 10-year cumulative weakest piece, etc.
duration of load effect (normal loading). 4.5.5 The nonparametric lower tolerance limit (NTL) of a
4.5 Statistics shall be shown with three significant digits. specified content is the mth order statistic, where m depends
Adequate significant digits shall be maintained in all interme- upon the sample size and confidence level. Table 2 depicts the
diate calculations to avoid rounding errors in the statistics. order statistic required to determine the lower-5 % NTL at a
4.5.1 The sample mean is calculated as follows: given sample size and three confidence levels. For example, if
n the sample size was 93 and the confidence level was chosen to
X 5 ( xi/n (5) be 95 %, m = 2. That is, the lower-5 % NTL with at least 95 %
i51
confidence would be the second order statistic. If other lower
where: percentiles are estimated, the corresponding NTLs can be
xi = individual observations, and determined (8, 12).
n = sample size. 4.5.6 If parametric methods are used, the parametric point
The sample mean is an unbiased estimator of the true estimate (PPE) and lower parametric tolerance limit (PTL)
population mean. shall be estimated by procedures documented as adequate for
4.5.2 The sample standard deviation is calculated as fol- the method adopted (1, 8, 12).
lows: 4.5.7 A histogram, or an empirical cumulative distribution

function, or both, shall be presented. The class widths for a


n
( ~xi 2 2 [~( xi! 2/n]
i51
histogram depend on the property; maximum widths are given
s5 n1 or (6) in Table 6. If parametric procedures are used for analysis,
either a cumulative distribution function or a probability


n
( ~xi X!2 density function can be shown superimposed on the empirical
i51
5 cumulative distribution function or the histogram respectively.
n1
NOTE 7Two examples of typical test data and a summary of the
results that meet the requirements of 4.1-4.5 are given in Appendix X1 and
4.5.3 The confidence interval (CI) for the mean is calculated Appendix X2.
as follows:
4.5.8 If a sampling unit other than an individual piece of
CI 5 X 6 ~ts/= n! (7) lumber is used, then the calculation of sample means, standard
where t depends on the sample size and confidence level, and deviations, confidence intervals, tolerance limits, and exclusion
is given in Table 1. A CI of this type provides that, if the limits must be made in a manner statistically consistent with
population is normally distributed, a given percent of all the sampling procedure chosen.
intervals found in this manner are expected to contain the true 4.6 If the purpose of the testing program is to evaluate the
population mean. accuracy of existing allowable properties for the population
4.5.4 The sample nonparametric percent point estimate sampled, this is done using the results of 4.5.3, 4.5.5, 4.5.6, or
(NPE) may be interpolated from the sample. To perform the 4.5.8. If an allowable mean property for a population falls
interpolation, arrange the test values in ascending order. within the confidence interval obtained in accordance with
Symbolically, call them x1, x2, x3, ... xn. Beginning with the 4.5.3, the testing program bears out the value allowed for the
lowest value (that is, first order statistic, see Note 6), calculate population with the associated confidence. The accuracy of
i/(n + 1), where i is the order of the value, for each successively existing near-minimum properties may be assessed using the
higher value until i/(n + 1) $ k/100, call it the jth value, equals results of 4.5.4, or 4.5.5, 4.5.6, and 4.5.8, or combination
or exceeds the sample k percentile point estimate. Interpolate thereof. If the existing property falls at or below the point
the nonparametric k percentage point estimate by: estimate as calculated in 4.5.4, the testing program may bear
out the existing values, but no confidence statement may be

TABLE 5 Reduction Factors to Relate Test Statistics to Allowable TABLE 6 Maximum Class Width to Be Used in Histogram Plots
Properties
Class Width, psi
Property
Property Factor (MPa)
Modulus of elasticity 1 Modulus of elasticity 100 000 (690)
Bending strength 1/2.1 Bending strength 500 (3.4)
Tensile strength 1/2.1 Tensile strength 500 (3.4)
Compressive strength parallel to grain 1/1.9 Compressive strength parallel to grain 500 (3.4)
Shear strength 1/2.1 Shear strength 50 (0.34)
Compressive strength perpendicular to grain 1/1.67 Compressive strength perpendicular to grain 50 (0.34)

6
D 2915 03
associated with this conclusion. In order to associate a confi- 5.2 Where properties have been previously assigned to a
dence statement, the existing value must fall below the lumber population, one purpose of this practice is to provide a
tolerance limit as calculated in 4.5.5, 4.5.6, or 4.5.8. format for evaluation of this assignment through full-size
4.7 If the purpose of the testing program is to establish lumber tests. Provisions are made for estimating both the mean
allowable properties for the population, this is done using the and near-minimum property values.
results of 4.5.1, 4.5.4, 4.5.5, 4.5.6, or 4.5.8. The allowable 5.3 Results obtained following the procedures and analyses
value of modulus of elasticity shall be the sample mean of of this practice may also be used to characterize the population
4.5.1, if the width of the confidence interval is a sufficiently
sampled for establishing design values. The specific character-
small fraction of the mean (for example, if ts/( X = n ) # l ,
ization with respect to the population, such as the mean or a
where l, predetermined by the user will normally be in the
near-minimum property, depends on the objective, the content,
range from 0.01 to 0.10). If this condition is not satisfied,
additional samples must be taken as described in 3.3 until the and confidence associated with the test sample. The represen-
condition holds. Generally, the allowable value of any near- tativeness and size of the sample influence how the character-
minimum allowable stress shall be the sample 5 % NPE of ization can be made. Contemporary practice is reflected in 4.7,
Section 4.5.4, if the relative difference between the NPE and however, other interpretations may be appropriate.
the NTL is sufficiently small, (that is, if (NPE NTL)/NPE < 5.4 The end use of a specific product will dictate the
d, where d will normally be in the range from 0.01 to 0.10). specification requirement. Indeed this practice addresses itself
This condition is essentially that of having sufficiently narrow to the procedures for sampling specified populations and
confidence interval for the NPE. If this condition is not procedures for interpreting the results. It cannot be imple-
satisfied, additional samples may be taken until the condition mented without the selection of values for the confidence levels
holds, or the NTL may be used for the allowable value. If the and degree of precision needed at various stages of the
latter course is chosen, one should be cognizant of the procedures. These values should be given careful consideration
imprecision in the NTL consequent on the sample size (see so that they are compatible with the anticipated end use, the
3.4.3.1). Alternatively, the PPE and PTL of the parametric risks that surround imprecise estimates, or incorrect decisions,
procedures provided for in 4.5.6 may be employed in a parallel and the costs of sampling, testing, and analysis.
manner.
5. Applications 6. Keywords
5.1 The results, reduced to the level of allowable properties, 6.1 lumber; structural lumber; wood
may be used to evaluate the accuracy of existing allowable
properties or to establish allowable properties.

APPENDIXES

(Nonmandatory Information)

X1. TYPICAL EXAMPLECOMMODITY LUMBER

X1.1 Population DescriptionSelected at random, from as an example of data that is typically recorded (Table X1.3).
one mill, were 80 No. 2 grade Hem-Fir two-by-fours (current It may be desirable to tabulate additional information, such as
lumber agency grade rules). The 80 test specimens were specific gravity, knot location, etc., depending on purpose.
equilibrated to an average of 15 % moisture content (see Note Note that tensile strength data is ordered in ascending order.
4 and Practice D 245).
X1.6 If appropriate best fit tests have been carried out and
X1.2 The purpose of the test was to evaluate the bending documented, only the best fit distribution need be illustrated;
modulus of elasticity, E, and tensile strength, F t, of a one-mill however, illustration of other options is instructive (see Table
sample relative to present design values. Consequently the fifth X1.4). Note that the nonparametric estimates in Table X1.4 for
percentile estimate will be considered for strength and the tensile strength can be estimated directly from Table X1.3, but
mean value for E (see 4.7). the estimates for modulus of elasticity are based on data, most
of which, is not shown in Table X1.3.
X1.3 The design value for the grade and species sampled is
given in Table X1.1. A table of test statistics is given in Table X1.7 Using 4.3, 4.5.1, 4.5.2, 4.5.3, and 4.6, the confidence
X1.2. interval for the mean E value (Table X1.2) did not contain the
X1.4 Histograms and fitted normal, lognormal, and Weibull value as printed in Table X1.1. Consequently, it was decided
distributions of edgewise bending E and tensile strength are this sample E did not verify the design E. Analysis of the
shown in Fig. X1.1 and Fig. X1.2. tension strength values was conducted in accordance with 4.5.4
and 4.5.5. After adjusting the nonparametric lower -5 %
X1.5 Several of the individual test results are shown, only tolerance limit to an allowable design value (that is, 1152/

7
D 2915 03

FIG. X1.1 Static Edgewise Modulus of Elasticity (106 psi)

FIG. X1.2 Tensile Strength (1000 psi)

2.1 = 548.6 psi (3.8 MPa)), it can be seen that this value is TABLE X1.1 Design Values for No. 2 Grade Hem-Fir Two-by-
below the value shown in Table X1.1 (Ft = 675 psi (4.6 MPa)); FoursA
therefore, the sample tension values do not verify the design Design Values
Species/Grade
Ft, psi (MPa) E, psi (MPa)
tension value. Hem-Fir No. 2 675 (4.6) 1 400 000 (9 646)
A
X1.8 Similar analyses could be performed using parametric National Design Specification for Wood Construction.

procedures and employing the values shown in Table X1.4.

8
D 2915 03
TABLE X1.2 Example Test Results for No. 2 Grade Hem-Fir Two-by FoursA
Property Mean, psi (MPa) Confidence Interval for Mean, psi (MPa)B Standard Deviation, psi (MPa) Sample Size
Static edgewise modulus of elasticityC 1 201 600(8 279) 1 148 500(7 113)1 254 700(8 645) 238 500(1 643) 80
Tensile strength 1 250(8.6) 1 100(7.6)1 350(9.3) 547(3.8) 80
A
All statistics in psi; all adjusted to 15 % moisture content in accordance with 4.2; reductions in accordance with 4.4 (not rounded).
B
95 % confidence.
C
Adjusted to ,/d of 21 and uniform load.

TABLE X1.3 Example of Test Results Ordered by Tensile StrengthTwo-by-Four Sample


Edgewise Modulus of Bending
Moisture Content Tensile Strength,
Specimen Number Elasticity 102, psi Width, in. (mm)B Thickness, in. (mm)B Strength
at Test, % psi (MPa)
(MPa)A Ratio, %C
1 P 43 15.0 1004 (6.9) 994 (6849) 3.47 (88) 1.47 (37) 13
1 P1 15.0 1092 (7.5) 959 (6607) 3.47 (88) 1.51 (38) 13
1 P 15 13.0 1152 (7.9) 1061 (7310) 3.42 (87) 1.50 (38) 52
1 P 28 15.0 1169 (8.0) 667 (4596) 3.45 (88) 1.46 (37) 47
1 P 22 16.0 1257 (8.7) 950 (6545) 3.46 (88) 1.49 (38) 52
A
Test ,/d of 44, quarter-point load, corrected to ,/d of 21 and a uniform load.
B
At test moisture content.
C
Obtained by 5.3.4.1 of Practice D 245.

TABLE X1.4 Estimates of Population Parameters for Two-by-Four


Sample
Static Edgewise
Tensile StrengthA, psi
Parameter Modulus Elasticity 106,
(MPa)
psi (MPa)
Weibull:
5 % point estimate 0.8255 (5688) 1.230 (8.5)
Lognormal:
5 % point estimate 0.8549 (5890) 1.270 (8.7)
5 % TL (75 %) 0.8340 (5746) 1.208 (8.3)
Normal:
Mean 1.2016 (8279) 2.616 (18.0)
Standard deviation 0.2385 (1643) 1.149 (7.9)
5 % point estimate 0.8091 (5575) 0.726 (5.0)
5 % TL (75 %) 0.7790 (5367) 0.580 (4.0)
Nonparametric:
5 % point estimate 0.8745 (6025) 1.169 (8.0)
5 % TL (75 %) 0.8490 (5850) 1.152 (7.9)
A
Not reduced to allowable property.

X2. TYPICAL EXAMPLELADDER RAIL STOCK

X2.1 Population Description(Species) ladder rail stock properties. Only mean and lower tail properties estimated by
graded in accordance with the (Grading Rules) as V.G. Ladder nonparametric procedures were of interest. The 95 % confi-
Rails. Two hundred pieces of 138 by 234-in. by 8 ft were dence level was deemed appropriate for both E and MOR in
selected randomly from stock at a ladder manufacturer in this study.
(location). Specimens were equilibrated in a conditioning
room. Actual average moisture content of specimens equaled X2.2 Data reduced to summary statistics are shown in Table
11.2 %. The standard deviation was 1.4 %. The purpose of the X2.1. Examples of individual specimen data are shown in
sampling, testing, and analysis was to obtain the bending Table X2.2; Table X2.3 contains estimates of near-minimum
modulus of rupture (MOR) and modulus of elasticity (E) of values. Histograms of test results are shown in Fig. X2.1 and
typical ladder rails for use in a research study on ladder rail Fig. X2.2. Empirical cumulative distribution functions are

TABLE X2.1 Ladder Rail Test StatisticsA


Confidence Interval Standard Deviation, Sample
Property Mean, psi (MPa)
for Mean, psi (MPa)B psi (MPa) Size
Static edgewise modulus 1 755 300 (12 094) 1 713 200 (11 804)1 797 400 (12 384) 301 500 (2077) 200
of elasticityC
Modulus of ruptureD 9 758 (67) 9 520 (66)10 014 (69) 1 836 (12.6) 200
A
All statistics in psi.
B
95 % confidence.
C
Adjusted to ,/d of 21, uniform load, and 12 % moisture content.
D
Adjusted to 12 % moisture content.

9
D 2915 03
TABLE X2.2 Sample Test ResultsLadder Rail
Moisture Content at Modulus of Rupture, Edgewise Static E, Width at Test, Thickness at Test,
Specimen
Test, % psi (MPa)A 106psi (MPa) A,B in. (mm) in. (mm)
103 12.8 14 343 (99) 2.51 (17 294) 2.753 (70) 1.366 (35)
111 11.0 11 423 (79) 1.80 (12 402) 2.760 (70) 1.381 (35)
114 8.6 6 505 (45) 1.37 (9 439) 2.784 (71) 1.406 (36)
121 11.6 9 708 (69) 2.17 (14 951) 2.762 (70) 1.386 (35)
A
Statistics adjusted to 12 % moisture content in accordance with 4.2; not adjusted to allowable properties.
B
Adjusted to ,/d of 21 and uniform load; actual conditions were ,/d of 33 and center point load.

TABLE X2.3 Estimates of Near-Minimum Population Parameters of Ladder Rail


5 % Tolerance Limits 5 % Tolerance Limits 5 % Tolerance Limits
Property 5 % Point Estimates
75 % Confidence 95 % Confidence 99 % Confidence
Edgewise modulus of 1.30 (8957) 1.29 (8888) 1.23 (8475) 1.16 (7992)
elasticityA
Modulus of ruptureB 6518 (45) 6072 (42) 5364 (37) 5353 (37)
A 6
10 psi (MPa); adjusted to ,/d of 21 and uniform load in accordance with 4.5.1; adjusted to 12 % moisture content in accordance with 4.2; not reduced to allowable
property.
B
psi (MPa); adjusted to 12 % in accordance with 4.2; not reduced to allowable property.

6
FIG. X2.1 Edgewise E (10 psi) FIG. X2.2 Bending Strength MOR (1000 psi)

shown in Fig. X2.3 and Fig. X2.4.


comparisons between the NPE and several NTLs can be made
X2.3 Following the procedures of 4.7 it was determined (Table X2.3). Maintaining the 10 % relative difference criterion
that the dispersion of E (static edgewise) measurements met the (NPE-NTL/NPE < 0.10) the relative difference for the NTL at
5 % requirement (that is, ts/ X = n = 0.024 # 0.05) with 95 %
a 95 % confidence level does not meet the criterion (6518-
confidence. Consequently, the research suggested an edgewise
5364/6518 = 0.17 > 0.10). Therefore, the 95 % confidence
E of 1.7 3 106 psi could be used as a design value (Practice
level goal of X2.1 for MOR is not met. Either more sampling
D 245 rounding rule would round the test value to 1.8 but this
would be out of the confidence interval for the mean, thus 1.7 (see 4.7) is required or the NTL (5364 psi (37 MPa)) may be
was chosen). used as the best estimate of the population MOR.

X2.4 Continuing the procedures of 4.7 for the MOR,

10
D 2915 03

FIG. X2.3 Empirical Cumulative Distribution Function


for E

FIG. X2.4 Empirical Cumulative Distribution Function


for R

X3. EXAMPLESAMPLING PROCEDURE

X3.1 When sampling from current production (that is, from therefore, 10k = 50. Select ten random numbers between 1 and
the green chain) at a manufacturing facility, the following 50. These are the ten random start points; 3, 9, 14, 29, 31, 36,
procedure allows the estimation of a standard error (SE) of the 40, 42, 47, and 50 (Table X3.1). Systemically select test
estimate as well as some information about the within-and- specimens using an interval length of 10k beginning at each of
between sample variance. the random start points (that is, random start x + 10k).
X3.2 Following the procedure outlined in Practice E 105
(A1.6) k is generally chosen to be five or greater. Let k = 5,

11
D 2915 03
TABLE X3.1 Test Specimens to Be SelectedA
3 9 14 29 31 36 40 42 47 50
53 59 64 79 81 86 90 92 97 100
103 109 114 129 131 136 140 142 147 150
153 159 164 179 181 186 190 192 197 200
... ... ... ... ... ... ... ... ... ...
... ... ... ... ... ... ... ... ... ...
... ... ... ... ... ... ... ... ... ...
A
This process is continued until the desired sample size is obtained.

X4. EXAMPLE OF USE OF TABLE 5 TO ADJUST MODULUS OF ELASTICITY (MOE) TO STANDARD


CONDITIONS

X4.1 An average apparent modulus of elasticity was h2/L2 5 1:21


obtained by testing simply supported beams loaded at the E 1 5 1.60 million psi
center and having a span-depth (L /h) ratio of 14:1. The MOE E/G 5 16
value obtained was 1.60 million psi. Assuming an E /G ratio of K1 5 1.20
16:1, what would be the apparent MOE for loads applied at the
K2 5 0.939
one-third points of the span with a span-depth ratio of 21:1?
Deflections were measured at the center of the span. Therefore,
From Table 4: E2 5 ~1.09796/1.034070!*1.60 million psi (X4.2)
h1/L1 5 1:14 (X4.1) E2 5 1.70 million psi

X5. ONE-SIDED TOLERANCE LIMITS FOR A NORMAL DISTRIBUTION

X5.1 A one-sided tolerance limit, PTL, is a value about g = (4n 5)/(4n 4), and
which it may be said with confidence 1-g, that at least a Zp and Zg are calculated with the following formula:
proportion, 1-p, of the population is greater than PTL. The
Z 5 T ~b0 1 b1T 1 b2T2!/~1 1 b3T 1 b 4T2 1 b5T3! (X5.3)
formula is as follows:
PTL 5
X Ks (X5.1) where:
where X and s are the mean and the standard deviation, T = =Ln~1/Q2! (Q = p for Zp and Q = g for Zg)
respectively, calculated from the sample data. K depends upon b0 = 2.515517
sample size n, as well as percentile 100-p and confidence 1-g. b1 = 0.802853
K values are given in Table 3 or they may be calculated from b2 = 0.010328
the following formula: b3 = 1.432788
b4 = 0.189269
Zpg 1 =Zp 2g 2 2 [g 2 2 Z g 2/~2~n 2 1!!#~Zp 2 2 Zg 2/n! b5 = 0.001308
K5
g 2 2 Zg 2/~2~n 2 1!!
NOTE X5.1K values computed using Eq X5.2 are approximations
(X5.2)
(see Ref (15)). For small values, the formula can seriously overestimate
where: the K factors.

REFERENCES

(1) Dodge, H. F., and Romig, H. G., Sampling Inspection Tables, Second tion, Biometrika, Vol 64, 1977, pp. 583588.
Edition, John Wiley and Sons, New York, NY, 1959. (7) Stephens, M. A., Tests of Fit for the Logistic Distribution Based on
(2) Anderson, T. W., and Darling, D. A., Asymptotic Theory of Certain the Empirical Distribution Function, Biometrika, Vol 66, 1979, pp.
Goodness-of-Fit Criteria Based on Stochastic Processes, Annuals of 591595.
the Mathematical Statistics, Vol 23, 1952, pp. 193212. (8) Natrella, M. G., Experimental Statistics, National Bureau of Stan-
(3) Anderson, T. W., and Darling, D. A., A Test of Goodness of Fit, dards Handbook , 91, U.S. Government Printing Office, 1963.
Journal of the American Statistical Association, Vol 49, 1954, pp. (9) Woodruff, R. S., Confidence Intervals for Medians and Other Position
765769. Measures, Journal of the American Statistical Association, Vol 47,
(4) Lewis, P. A. W., Distribution of the Anderson-Darling Statistic, 1952, pp. 635646.
Annuals of the Mathematical Statistics, Vol 32, 1961, pp. 118121. (10) Johnson, R. A., and Haskell, J., Estimation of Tolerance Intervals
(5) Stephens, M. A., EDF Statistics for Goodness of Fit and Some and Selection Procedures for Property Characterization, United
Comparisons, Journal of the American Statistical Association, Vol 69, States Department of Agriculture, Forest Service, Interim Report,
1974, pp. 730737. Forest Products Laboratory, Madison, WI, 1978.
(6) Stephens, M. A., Goodness of Fit for the Extreme Value Distribu- (11) Hilbrand, H. C., Graphical Solution for Shear Deflection of

12
D 2915 03
Flexural Members, Forest Products Journal, Vol 17, No. 6, 1967, Report on file at ASTM Headquarters, 1985.
pp. 6162. (14) Link, Carol L., An Equation for One-Sided Tolerance Limits for
(12) Guttman, I. Statistical Tolerance Regions: Classical and Bayesian, Normal Distributions, Research Paper FPL 458, Madison, Wiscon-
Hafner Publishing Co., Darien, CT, 1970. sin, U.S. Department of Agriculture, Forest Service, Forest Products
(13) Palka, L. C., and Barrett, J. D. An Examination of E/G Values for Laboratory, 1985.
Canadian Spruce Lumber. Report to ASTM task group investigating (15) Guttman, Irwin, Statistical Tolerance Regions: Classical and Baye-
the validity of Table 2 of ASTM D2915 74. Forintek Canada Corp. sian, Hafner Publishing Co., Darien, CT, pp. 8893, 1970.

ASTM International takes no position respecting the validity of any patent rights asserted in connection with any item mentioned
in this standard. Users of this standard are expressly advised that determination of the validity of any such patent rights, and the risk
of infringement of such rights, are entirely their own responsibility.

This standard is subject to revision at any time by the responsible technical committee and must be reviewed every five years and
if not revised, either reapproved or withdrawn. Your comments are invited either for revision of this standard or for additional standards
and should be addressed to ASTM International Headquarters. Your comments will receive careful consideration at a meeting of the
responsible technical committee, which you may attend. If you feel that your comments have not received a fair hearing you should
make your views known to the ASTM Committee on Standards, at the address shown below.

This standard is copyrighted by ASTM International, 100 Barr Harbor Drive, PO Box C700, West Conshohocken, PA 19428-2959,
United States. Individual reprints (single or multiple copies) of this standard may be obtained by contacting ASTM at the above
address or at 610-832-9585 (phone), 610-832-9555 (fax), or service@astm.org (e-mail); or through the ASTM website
(www.astm.org).

13